diff options
Diffstat (limited to 'ChangeLog-99b')
| -rw-r--r-- | ChangeLog-99b | 2137 |
1 files changed, 4 insertions, 2133 deletions
diff --git a/ChangeLog-99b b/ChangeLog-99b index ef758e7cd65..988a5d01c46 100644 --- a/ChangeLog-99b +++ b/ChangeLog-99b @@ -1,2138 +1,9 @@ -Sun May 23 11:33:07 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - Changes for the following enhancement in the Proactor - implementation. - Enhanced the semantics of the <end_event_loop>, so that when it - is called, it wakes up all the threads blocked on waiting for - completions and end the event loop. - - Thanks to "Edwin D. Windes" <ewindes@usa.net> for reporting the issue. - - * ace/Proactor.h: - * ace/Proactor.cpp: - Added static helper methods and states to post the wakeup - completions to the CompletionPort. - - * ace/Asynch_IO.h: - * ace/Asynch_IO.cpp: - Added a method <handle_wakeup> to the ACE_Handler. - - * ace/Object_Manager.cpp - * ace/Object_Manager.h - - Added a new Thread_Mutex called <ACE_PROACTOR_EVENT_LOOP_LOCK> to - protect the thread count in the event loop. - - * ace/POSIX_Asynch_IO.cpp - * ace/POSIX_Proactor.cpp - * ace/POSIX_Proactor.h - * ace/Proactor.cpp - * ace/Proactor.h - * ace/Proactor_Impl.h - * ace/WIN32_Proactor.cpp - * ace/WIN32_Proactor.h - - Changes for posting wakeup completions. - - * examples/Reactor/Proactor/test_end_event_loop.cpp: - * examples/Reactor/Proactor/test_end_event_loop.dsp: - * examples/Reactor/Proactor/post_completions.dsp: - * examples/Reactor/Proactor/Makefile: - - Test file the <end_event_loop> enhancement. - -Sat May 22 16:57:38 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/FlReactor.cpp: - Sun/CC doesn't like reinterpret cast when converting from an int - into an int; this is silly, but I cannot use static_cast because - on NT reinterpret_cast is the right thing. - Use old-style casts for the moment. - - * tests/FlReactor_Test.cpp: - Minor cosmetic fixes. - -Sat May 22 16:36:14 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ACE version 4.6.40 released. - -Sat May 22 14:53:47 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Filecache.cpp: snagged another "template <>". - Thanks to Bala for reporting it. - -Sat May 22 07:30:44 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-chorus.h,ace/config-lynxos.h: added - ACE_HAS_NONCONST_SELECT_TIMEVAL. - - * ace/Filecache.cpp: commented out "template<>" for - now. Not all compilers can cope with it. - -Fri May 21 21:12:37 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Map_Manager.i (operator==): Since <map_man_> is a pointer - now, we must compare by value, rather than by address. - -Fri May 21 12:56:38 1999 David L. Levine <levine@cs.wustl.edu> - - * examples/IPC_SAP/SOCK_SAP/CPP-inclient.cpp: fixed to - build without threads: #ifdef ACE_MT_SAFE is insufficient. - Must also check if it is 0. - -Fri May 21 12:45:32 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Map_Manager.i: - Fixed minor error in previous commit. Thanks to Irfan for - helping out with this. - -Fri May 21 11:27:44 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Map_Manager.h (class ACE_Map_Iterator_Base): Changed the - ACE_Map_Manager reference to a pointer. This way there is no - problem in reinitializing <map_man_>, and hence the compiler - will not have a problem in providing us a default assignment - operator. Thanks to Knut Johannessen <knutj@funcom.com> for - pointing out this problem. - - * ace/Process: A nifty ACE::fork was added several months ago - which offers the option to avoid zombies, but it is not used in - ACE_Process::spawn. - - Mark L. Boriack <mboriack@sito.saic.com> provided patches which - adds an <avoid_zombies_> data member to ACE_Process_Options, - appropriate accessor functions, and changes the ACE_OS::fork - call to ACE::fork in ACE_Process::spawn, making use of the - option. The default behavior is left unchanged, i.e., you get - zombies. Thanks Mark. - -Fri May 21 10:34:10 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/gethrtime.cpp,Makefile (gethrtime on Green - Hills/VxWorks): reverted to building gethrtime.cpp with g++, because - the ghs assembly code didn't work properly. - -Fri May 21 10:10:13 1999 Ossama Othman <othman@cs.wustl.edu> - - * ACE-INSTALL.html: Updated information about building egcs' C++ - shared libraries with an example, and emphasized the fact that - the `--enable-shared' configuration option is not an egcs - run-time option. Thanks to Wallace Owen <owen@visicom.com> for - motivating this. - -Fri May 21 09:15:08 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/config-sunos5.5.h: - Added ACE_HAS_NONCONST_SELECT_TIMEVAL here too - -Fri May 21 07:53:58 1999 Steve Huston <shuston@riverace.com> - - * ace/config-vxworks5.x.h, config-osf1-4.0.h: Added - ACE_HAS_NONCONST_SELECT_TIMEVAL. - -Thu May 20 18:01:24 1999 Steve Huston <shuston@riverace.com> - - * ace/Filecache.cpp: Added "template<>" to specializations. - - * ace/OS.i (ACE_OS::select, ACE_OS::sleep): Removed implicit - dependency on binary object layout with regard to using an - ACE_Time_Value object as a struct timeval; use the timeval* - operator instead. On platforms with ACE_HAS_NONCONST_SELECT_TIMEVAL - don't just cast away const-ness of the ACE_Time_Value - modification - of the timeval by select() is not consistent, and Linux is not the - only one that modifies it (Solaris 7 does also, as well as AIX and - possibly HP; Solaris 2.5/6 don't specify). So, play it safe. - Inlining and optimization should prevent this from having any - affect on performance. - - * ace/config-sunos5.5.h, config-hpux-10.x.h, config-aix-4.x.h: - Added #define ACE_HAS_NONCONST_SELECT_TIMEVAL. - -Thu May 20 17:47:35 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/OS.h (ACE_DES_ARRAY_FREE_*): Added new macros to facilitate - proper destruction of arrays. - - * ace/Containers_T.cpp (ACE_Array_Base): Changed ACE_Array_Base to - use the new ACE_DES_ARRAY_FREE_* macros in order to prevent the - memory leaks. - -Thu May 20 17:46:02 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ace_dll.dsp: - * ace/ace_lib.dsp: Fixed Alpha configurations. Thanks to - Christian Schuderer <Christian.Schuderer@erl9.siemens.de> for - reporting this. - -Thu May 20 15:35:23 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * tests/TkReactor_Test.cpp: - It was not compiling if ACE_HAS_TK was not defined. - -Thu May 20 15:33:21 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_vxworks5.x_ghs.GNU: - added (and default to) LIBS for Green Hills 1.8.9. - -Thu May 20 14:34:59 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5.h: with Green Hills only, added - ACE_LACKS_ACE_IOSTREAM. ACE's IOStream_Test never halts - when built with Green Hills 1.8.9. - -Thu May 20 14:21:44 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i: (t_getname): fixed comment after #endif, to - agree with the #ifdef. - -Thu May 20 13:51:20 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/config-win32-common.h: I forgot to put '/**/' between - #include and <blah.h> which caused MSVC to generate warnings - when checking file dependencies. Thanks to Christian Schuderer - <Christian.Schuderer@erl9.siemens.de> for reporting this. - -Thu May 20 13:52:04 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-sunos5.5.h: build with threads if - _POSIX_C_SOURCE >= 199506L or #ifdef _POSIX_PTHREAD_SEMANTICS. - Added #define ACE_LACKS_RWLOCK_T if _POSIX_PTHREAD_SEMANTICS - is #defined (and therefore ACE_HAS_STHREADS is not #defined). - The rwlock functions are a Solaris threads interface, and - are not POSIX. - - * include/makeinclude/platform_sunos5_ghs.GNU: default to - building without threads support, because Green Hills 1.8.9 - is incompatible with the SunOS threads libraries. - -Thu May 20 12:12:01 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.{h,cpp},Sched_Params.cpp: with ACE_HAS_PRIOCNTL, - moved #include of sys/{rt,ts}priocntl.h from OS.cpp and - Sched_Params.cpp to OS.h, so that USYNC_THREAD and USYNC_PROCESS - #defines can be protected effectively. - -Thu May 20 11:19:45 1999 Joe Hoffert <joeh@cs.wustl.edu> - - * ace/ATM_Addr.cpp: Added ACE_UNUSED_ARG for selector parameter - when appropriate. - -Thu May 20 09:57:57 1999 Bill Rizzi <rizzi@softserv.com> - - * ace/Log_Msg.{h,cpp},examples/Log_Msg/test_log_msg.cpp - (priority_mask): added optional support for process priority - mask, in addition to the (now default) thread priority mask. - -Thu May 20 09:42:54 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h (fcntl): changed type of last argument from int to - long, to permit safer casting from pointer types. - - * ace/OS.i (flock_*): use ACE_OS::fcntl instead of ::fcntl. - - * ace/OS.cpp (ftruncate): use ACE_OS::fcntl instead of ::fcntl. - - Thanks to Joe Golio <Joe.Golio@xylan.com> for reporting that - these functions weren't using ACE_OS::fcntl. - -Wed May 19 23:50:18 1999 Nagarajan Surendran <naga@cs.wustl.edu> - - * ace/TkReactor.{h,cpp}: - * ace/Makefile: - Added a new reactor implementation based on the Tcl/Tk Event - loop and event handling mechanism. - - * ace/tests/TkReactor_Test.{cpp,tcl}: - * ace/tests/Makefile: - The Tk Reactor test and the tcl script for the GUI. This is - similar to the XtReactor_Test. - - * ace/include/makeinclude/wrapper_macros.GNU: - * ace/include/makeinclude/platform_sunos5_sunc++.GNU - Added changes to Makefile flags for XtReactor. - -Wed May 19 21:56:49 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Message_Queue.h: fixed typo in comment: interrupt. - -Wed May 19 21:34:29 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_4.0.GNU: removed these - warning suppressions: - 174: expr_has_no_effect - 610: nonoverriding_function_decl - They're certainly not needed with cxx 6.2-020, and - hopefully not needed with earlier versions. - -Wed May 19 10:52:15 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * examples/DLL/test_dll.cpp (main): Fixed auto pointer use. It - was causing "Temporary used for non-const reference" warnings. - -Wed May 19 10:04:54 1999 Jeff Parsons <parsons@cs.wustl.edu> - - * ace/SString.i: - Changed the default constructor to set rep_ to the - empty string instead of the null string. Thanks to - Knut Johannessen <knutj@funcom.com> for pointing - this out. - -Wed May 19 06:34:58 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-minimal.h: added #define ACE_USE_THREAD_MANAGER_ADAPTER, - so that ACE_Thread_Exit isn't needed in the ACE_OS layer. - -Tue May 18 22:03:49 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_4.0.GNU,platform_osf1_cxx.GNU: - removed suppression of warning 9 (nested comment not allowed) on - Digital Unix 5.0 and later. - -Tue May 18 15:01:39 1999 Steve Huston <shuston@riverace.com> - - * ace/ACE.cpp: Added #include "ace/SString.h" to catch def for - ACE_Tokenizer in Winsock 1.1 mode. - -Tue May 18 13:30:55 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * include/makeinclude/rules.nested.GNU: FOR is not an excutable on - Win32, therefore, we must bring in the command processor in to - use FOR. Thanks to David Hooker <DHooker@uniview.net> for - reporting the bug. - -Tue May 18 11:46:55 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/FILE_Connector.h: Changed the default perms from 0 to - ACE_DEFAULT_FILE_PERMS. - -Tue May 18 11:28:32 1999 Joe Hoffert <joeh@cs.wustl.edu> - - * examples/IPC_SAP/TLI_SAP/CPP-ATM-server.cpp: forced - ACE_TLI_Stream's recv method to call ACE_OS::t_rcv rather than - ACE_OS::read to be compatible with XTI. - -Tue May 18 11:00:15 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Task_T: Moved the ACE_Buffered_Task logic out of the Task - abstraction since it doesn't belong there. - - * ace/Makefile: Regenerated the dependencies. - - * ace/FlReactor.cpp: We should be including <FL/Fl.h> rather than - "FL/Fl.h". - -Tue May 18 10:42:47 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_4.0.GNU: removed - suppression of warning 1180, statement causes unreachable - return, because it's no longer necessary. - -Tue May 18 10:38:53 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Thread_Manager_Test.cpp: replaced Hash_Map_Manager - with a simple array. This avoids dynamic allocation in - the signal handler. And, it avoids having to specialize - ACE_Hash<ACE_hthread_t> on platforms that have a pthread_t - struct. - -Tue May 18 09:33:46 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/FILE_IO: Added new sendv()/recvv() and sendv_n()/recvv_n() - methods to be consistent with SOCK_IO and SOCK_Stream. - - * ace/SOCK_IO.cpp (recvv): Make sure to 0-out the iov_base so that - it's always ok to delete it, even if any of the internal - operations fail! - -Tue May 18 10:14:47 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Makefile: - * ace/FlReactor.h: - * ace/FlReactor.i: - * ace/FlReactor.cpp: - * include/makeinclude/platform_linux_lxpthread.GNU: - * include/makeinclude/wrapper_macros.GNU: - * tests/Makefile: - * tests/FlReactor_Test.cpp: - Added new reactor implementation based on the event loop of the - Fast-Light toolkit, similar to the XtReactor. - FLTK is a GUI toolkit that works on UNIX/X-Windows and on Win32, - it is integrated to OpenGL and is pure C++. - Check http://fltk.easysw.org for more details. - -Tue May 18 09:35:00 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Thread_Manager_Test.{h,cpp},Makefile: - removed Thread_Manager_Test.h. - -Mon May 17 22:39:06 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Thread_Manager_Test.cpp: added a couple more - explicit template instantiations. Thanks to Carlos - for reporting that they're needed on Irix. And, - added an ACE_Hash<ACE_hthread_t> specialization to - keep DU cxx happy. - -Mon May 17 16:31:16 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Task_T.cpp, - ace/OS.h: Removed ACE_DEFAULT_WRITEV_MAX and replaced it - with IOV_MAX. - -Mon May 17 15:49:48 1999 David L. Levine <levine@cs.wustl.edu> - - * config-dgux-4.x-ghs.h,config-vxworks5.x.h,config-ghs-common.h: - factored ACE_LACKS_SIGNED_CHAR to config-ghs-common.h. - - * config-dgux-4.x-ghs.h: removed ACE_LACKS_ACE_IOSTREAM. - It should work now, though we don't have that platform - to verify it. - -Mon May 17 15:08:45 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/CORBA_macros.h: - Fixed implementation of the ACE_READ_GUARD_THROW_EX and - ACE_WRITE_GUARD_THROW_EX macros. - -Mon May 17 15:12:54 1999 David L. Levine <levine@cs.wustl.edu> - - * test/Thread_Manager_Test.cpp: use a hash map, instead of - TSS, to store each thread's indication of whether it has - been signalled. That way, the signal handler doesn't have - to access TSS. Increased the delay from 1 to 2 seconds after - the signal test, which also seems to help avoid thread lib - panics on Solaris. - -Mon May 17 14:02:01 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Conn_Test.cpp (handler): added comment saying that even - the ACE_OS::exit () call isn't safe. - -Mon May 17 10:31:04 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Service_Repository.cpp (find_i): A missing pointer check was - causing the remove method to segfault. Thanks to Tom Arbuckle - <arbuckle@uran.informatik.uni-bonn.de> for pointing this out. - -Mon May 17 10:21:29 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/TSS_Test.cpp: removed the signal handler. It - wasn't used for the test, and contained an ACE_DEBUG. - -Mon May 17 10:12:45 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-chorus.h,config-dgux-4.x-ghs.h,config-sunos5.5.h, - config-vxworks5.x.h,config-ghs-common.h: - factored out Green Hill-specific #defines to config-ghs-common.h. - -Mon May 17 09:07:41 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Conn_Test.cpp (handler): commented out unused arg - "signum". - -Sun May 16 20:38:58 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/CDR_Stream.{h,i} (current): Added this accesor function for - Flick IDL compiler. Thanks to Tim Stack <stack@aros.net> for - recommending this. - -Sun May 16 20:08:41 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Thread_Manager_Test.cpp (handler,worker): - moved ACE_DEBUG out of signal handler. It was - causing occasional lockups on SunOS. Signal - handlers must never do anything that could cause - a lock to be acquired. - - * tests/Conn_Test.cpp (handler): removed ACE_DEBUG. - -Sun May 16 20:04:09 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: moved #include of ace/Trace.h to before #include of - ace/OS.i. - - * ace/OS.{i,cpp}: enabled ACE_TRACE for most functions (except - those that are used by ACE_TRACE). - -Sun May 16 19:54:10 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * ace/config-sunos5.7.h: Comments - * ace/config-sunos5.6.h: Comments - - * ace/POSIX_Asynch_IO.cpp: - * ace/POSIX_Proactor.h: - * ace/POSIX_Proactor.cpp: - Using ACE_dynamic_cast to get - <ACE_POSIX_Asynch_Result *> from <aiocb *> gives build errors on - egcs. <ACE_reinterpret_cast> is not the correct solution for - this. Solved this problem by having one more list of - <ACE_POSIX_Asynch_Result *> so that we dont have to do any - casting. - * ace/Proactor.cpp (ACE_Proactor): Indentation. - * examples/Reactor/Proactor/post_completions.cpp: Added template - pragma's. - -Sun May 16 18:35:59 1999 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.6.39 released. - -Sun May 16 12:34:48 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Log_Msg.cpp (ACE_Log_Msg_Manager::lock_): changed - the lock type from ACE_Thread_Mutex to - ACE_Recursive_Thread_Mutex. This allows ACE_Log_Msg's - methods to be called after calling ACE_Log_Msg::acquire (), - without deadlock. Thanks to Bill Rizzi <rizzi@softserv.com> - for suggesting this. - -Sat May 15 08:30:43 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Containers.h: added #include of ace/Malloc_Base.h. - It's needed for the definition of ACE_Allocator when - instantiating container classes. - -Sat May 15 03:32:02 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/CDR_Stream.h: Added #include of ace/SString.h. - -Fri May 14 23:05:32 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Log_Record.h: removed unnecessary #include of ace/SString.h. - -Fri May 14 21:11:46 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Task_T.h: - * ace/Task_T.cpp: - Fixed syntax errors in ACE_Buffered_Task - -Fri May 14 21:03:30 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * netsvcs/lib/Log_Message_Receiver.cpp: - Changed ACE_Guard instantiations with ACE_GUARD() macros to - avoid some warnings with egcs, similar to the problem that David - had with Message_Queue_T.cpp - -Fri May 14 19:31:04 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/Message_Block.cpp (total_size): - - Corrected warning caused by assignment to non-const from const. - - * ace/Message_Queue_T.cpp (close): - - Temporary variable wasn't declared. - - * ace/Message_Queue_T.cpp (remove_messages): - - Removed unused variable. - -Fri May 14 19:16:06 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/Task_T.cpp (put, flush): - - Corrected syntax errors where return type was placed before - template keyword. - -Fri May 14 16:46:42 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Task_T: Added a new class called ACE_Buffered_Task that - buffers up its ACE_Message_Blocks until (1) the buffer is full - or (2) a timeout elapses, at which point the buffers are written - out using some type of writev()-like mechanism. - - * ace/OS.h: Added a new macro called ACE_DEFAULT_WRITEV_MAX, which - defaults to the maximum value that writev() can have. - - * ace/Message_Block.cpp: Added a new method called total_size() - that returns the total number of bytes in a composite - Message_Block (i.e., a Message_Block that has a chain of - Message_Blocks). - - * ace/Message_Queue_T.cpp: Reimplemented all the hand-coded loops - to use the new ACE_Message_Block::total_size() method. - -Fri May 14 16:17:27 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Synch.h (ACE_Null_Mutex>): Conditionally compiles ~ACE_Guard() - destructor. Win32 needs this definition. - -Fri May 14 16:10:29 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Message_Queue_T.cpp: replaced raw ACE_Read_Guards - with ACE_READ_GUARD_RETURN macros. That avoids a compiler - warning (egcs 1.1.1/SunOS 5.7) now that the ACE_Guard<Null_Mutex> - destructor has been removed. Thanks to Irfan and Carlos - for suggesting this fix. - -Fri May 14 15:35:50 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/ACE.cpp (clr_flags and set_flags): We should use ioctl() - calls to implement on vxworks this rather than fcntl(). Thanks - to Joseph A. Condlin <jac@slpmbo.ed.ray.com> for reporting this. - -Fri May 14 13:15:02 1999 Joe Hoffert <joeh@cs.wustl.edu> - - * examples/IPC_SAP/TLI_SAP/Makefile: Took out FORE specific - information. Somehow my local changes got checked in - this is - to undo that. - -Fri May 14 12:58:02 1999 Joe Hoffert <joeh@cs.wustl.edu> - - * ace/ATM_Addr.h: - * ace/ATM_Addr.i: - * ace/ATM_Addr.cpp: Added accessor methods for the selector byte - of an ATM address. - - * examples/IPC_SAP/TLI_SAP/CPP-ATM-client.cpp: - * examples/IPC_SAP/TLI_SAP/CPP-ATM-server.cpp: Modify example code - to have the user optionally specify the selector byte. - -Fri May 14 12:47:38 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Addr.cpp,ace/Synch_Options.cpp: added comments about - the static objects. - -Fri May 14 11:36:05 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/SString.h: - * ace/SString.i: Added new class to auto destroy dynamically - allocated strings, sort of a mixing between auto_ptr and - CString. Auto_Ptr_Array was not useful because we need to free - using ACE_OS::free() (and in the future allocators). - -Fri May 14 10:37:49 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/OS.i: Fixed a deadlock in cond_timedwait()/cond_broadcast(). - Thanks to Brian C. Olson <bolson@rtlogic.com> for reporting - this. - - * ace/Synch.h (ACE_Null_Mutex>): Commented out the ~ACE_Guard() - destructor to prevent segfaults on certain versions of SunC++. - Thanks to Russ Noseworthy for reporting this bizarre bug! - -Fri May 14 09:51:55 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Message_Block.h: - * ace/Message_Block.cpp: - Modified default constructor and constructor taking a Data_Block - to also take an Allocator. - -Fri May 14 09:14:55 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_cxx.GNU: - added $(ACELIB) to SOFLAGS for DU 5.x. It should - have been there all along. - -Thu May 13 17:27:14 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * netsvcs/servers/svc.conf: - * netsvcs/clients/Naming/Client/svc2.conf: - * netsvcs/clients/Naming/Client/svc2.conf: Removed relative paths - from all DLL's. They were hard-coded which was not portable. - ACE should figure out the correct path automatically. - - * netsvcs/lib/Name_Handler.cpp: Replaced several calls to - ACE_WString::rep with ACE_WString::fast_rep to avoid memory - leaks. Thanks to Valery Arkhangorodsky - <valerya@servicesoft.com> for reporting this. - - * ace/Registry.cpp (close): Only close <key_> when we do own a - key. - - * netsvcs/lib/Name_Handler.cpp (resolve): The atype string needs - to be delete[]'ed once we are done with it. Thanks to Valery - Arkhangorodsky <valerya@servicesoft.com> for pointing this out. - -Thu May 13 09:39:17 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Makefile: added Tokens_Test to BIN2 if Tokens component - is not configured in. This ensures that Tokens_Test gets - cleaned up, even if the ace directory was realcleaned first - (and ACE_BUILD_COMPONENTS is therefore empty). - -Thu May 13 07:24:41 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/POSIX_Proactor.cpp (handle_events): replaced - incorrect dynamic_cast with a reinterpret cast. - -Wed May 12 17:42:09 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Message_Block.h: - * ace/Message_Block.cpp: - The message blocks can be obtained from an allocator now. Thanks - to John Aughey <Aughey@concentric.net> for this cool feature. - -Wed May 12 16:24:17 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * examples/Reactor/Proactor/Makefile: - * examples/Reactor/Proactor/test_proactor.cpp: - * ace/POSIX_Proactor.cpp: - * ace/POSIX_Proactor.h: - * ace/Proactor.cpp : - * ace/POSIX_Asynch_IO.cpp: Fixed the problem with terminating the - Timer thread. Also, fixed the places where conversion between - void * and ACE_POSIX_Asynch_Result. Used ACE_reinterpret_cast at - these places. - -Wed May 12 12:14:09 1999 Ossama Othman <othman@cs.wustl.edu> - - * COPYING: Additional updates/clarifications suggested by Richard - Stallman. - -Wed May 12 00:25:57 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Thread_Manager.h (ACE_Thread_Manager): Added clarification - for the semantics differences among wait(), wait_grp() and - wait_task(). - -Tue May 11 18:11:51 1999 James C. Hu <jxh@entera.com> - - * ace/config-sunos5.7.h: SunOS 5.7 (aka Solaris 7) does not have a - limited select. - -Tue May 11 17:11:51 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Conn_Test.cpp (spawn_threads): changed some ACE_NEW - statements to ACE_NEW_RETURN. - -Tue May 11 15:43:30 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: added ACE_THR_PRI_OTHER_DEF for NT, because MT_Cubit - uses it. - -Tue May 11 14:45:08 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Conn_Test.cpp: fixed use of status on platforms that lack - fork. - -Tue May 11 14:10:29 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-osf-4.0.h: added #define of - ACE_NEW_THROWS_EXCEPTIONS, if exception handling is enabled. - - * tests/New_Fail_Test.cpp: removed workaround for DEC cxx, and - modified error message to mention ACE_NEW_THROWS_EXCEPTIONS. - -Tue May 11 12:02:03 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-win32-common.h: set __ACE_INLINE__ to 1 if not - #defined. It's scary when I trip over syntax errors on Win32 - :-) - -Tue May 11 10:00:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/OS.i: fixed name of unused argument for pSOS - - * ace/RB_Tree.{cpp, h}: moved enumerated type from template to - non-templatized base class to make old g++ happy. - -Mon May 10 22:59:46 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/wrapper_macros.GNU: don't set VDIR to - .shobj/ for VxWorks, because it uses the same .o files for - static and shared libs. - - * include/makeinclude/platform_vxworks5.x_g++.GNU: disable RTTI - for egcs, because the Tornado 1.0.x libs don't support it. No - harm in disabling for old g++. - -Mon May 10 15:33:56 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/INET_Addr.cpp (string_to_addr): Fixed a mistake that - occurred when integrating Jerry D. De Master - <jdemaste@ipdinc.com> fix. Thanks to Zoran and Jerry for - reporting this. - -Mon May 10 15:00:31 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ACE-INSTALL.html: Changed to use ACE_NO_INLINE to supress - inlining to match with the recent dsp file changes. Added a - link to Barry Hoggard's CORBA page which contains ACE related - MSVC setting hints. - - * ace/Select_Reactor_Base.cpp (unbind): We need to make sure the - reactor's suspend_set_ also gets cleaned up when unbinding a - handler. Thanks to Casey Lucas <clucas@sabre.com> for pointing - this out. - -Mon May 10 17:08:29 1999 James C. Hu <jxh@entera.com> - - * ace/ACE.cpp: - * ace/config-sunos5.6.h: - * ace/README: Added support to enter_recv_timedwait, - enter_send_timedwait, handle_timed_complete, and - handle_timed_accept to use poll() instead of select() if poll() - is available and select() is known to have problems. Created - ACE_HAS_LIMITED_SELECT macro. Certain versions of SunOS 5.6 do - not correctly select on file descriptors that are larger than - 1023. - -Mon May 10 13:32:11 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Conn_Test.cpp (spawn_threads): wait a maximum of 200 - seconds. Carlos reported that the test sometimes hangs on Irix. - -Mon May 10 12:14:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.cpp: fixed a dangling pointer bug in - RB_Tree::delete_fixup. Added pointer tests to all dereferences - in this piece of code. - -Mon May 10 11:35:18 1999 Ossama Othman <othman@cs.wustl.edu> - - * COPYING: Made some updates suggested by Richard Stallman - <rms@gnu.org> of the Free Software Foundation - (http://www.fsf.org/). - -Mon May 10 11:01:58 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/New_Fail_Test.cpp: DEC cxx catches its default, - NoNamedException. OK, I guess. - - * tests/Upgradable_RW_Test.cpp: added -f option, which enables - FIFO thread scheduling. Some platforms require superuser - privilege to use it. Also, wait a maximum of 1 second per - iteration. And, replaced some u_longs with u_ints. - - * tests/Thread_Manager_Test.cpp (main): wait a maximum of 60 - seconds, because the test sometimes hangs on SunOS 5.5.1 and - SunOS 5.7. - -Mon May 10 08:26:44 1999 David L. Levine <levine@cs.wustl.edu> - - * bin/g++dep: added support for our Solaris 7 machines. - -Sun May 9 18:07:24 1999 Ossama Othman <othman@cs.wustl.edu> - - * acconfig.h: - * configure.in: Added support/tests for - ACE_LACKS_CHAR_RIGHT_SHIFTS and - ACE_LACKS_CHAR_STAR_RIGHT_SHIFTS. - - * ace/README: Fixed typo. ACE_LACKS_LINEBUFFERED_STREAMBUF was - used twice. Corrected the appropriate one to - ACE_LACKS_UNBUFFERED_STREAMBUF. - -Sun May 09 11:45:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.cpp added explicit scoping to RB_SearchResult - parameter in RB_Tree<...>::find_node (). - -Sun May 09 11:30:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Makefile - examples/Connection/non_blocking/Makefile - examples/IPC_SAP/SPIPE_SAP/Makefile - examples/IPC_SAP/TLI_SAP/Makefile - examples/Reactor/Misc/Makefile - examples/Service_Configurator/IPC-tests/client/Makefile - examples/Service_Configurator/IPC-tests/server/Makefile - netsvcs/lib/Makefile - netsvcs/servers/Makefile - tests/Makefile: ran make depend after adding new RB_Tree - interface. - -Sat May 08 19:30:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.{cpp, h, i} - tests/RB_Tree_Test.cpp: finished implementing all methods for - the ACE_Hash_Map_Manager_Ex compliant interface, fixed a bug in - the internal tree search method (thanks to David Levine for - pointing this out), and added a new section to the test that - exercises the new bind and unbind methods. - -Sat May 08 17:09:10 1999 David L. Levine <levine@cs.wustl.edu> - - * apps/Gateway/Gateway/Event_Channel.cpp (bind_proxy,subscribe): - wrapped final return statement with ACE_NOTREACHED, to please DU - cxx 6.2. - -Sat May 08 08:49:44 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/CDR_Stream.h: only use unsigned long for Boolean with Green - Hills 1.8.8 on Chorus, not 1.8.9. - - * ace/IOStream.h,README: added ACE_LACKS_CHAR_STAR_RIGHT_SHIFTS - and ACE_LACKS_CHAR_RIGHT_SHIFTS support. - - * ace/config-chorus.h: added ACE_LACKS_CHAR_STAR_RIGHT_SHIFTS - #define for g++. - - * ace/config-vxworks5.x.h: added ACE_LACKS_CHAR_RIGHT_SHIFTS - #define for Green Hills 1.8.9. - -Sat May 08 08:44:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.{cpp, h, i} - tests/RB_Tree_Test.cpp - apps/JAWS/PROTOTYPE/HTTPU/http_headers.cpp: Added more - implementation code and necessary template instantiations for - migration of RB_Tree class templates to conform to the - ACE_Hash_Map_Manager API. - -Sat May 08 03:59:52 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/config-win32-common.h: Re-ordered __ACE_INLINE__ macro check - and definition so users can turn of inlining correctly. - However, the right thing to do is to define ACE_NO_INLINE macro. - - * ace/ace_{dll,lib}.dsp: Moved RB_Tree.cpp to template folder and - excluded it from being built. - - * ace/ace_lib.dsp: - * performance-tests/Misc/basic_perf.dsp: - * performance-tests/Misc/context_switch_time.dsp: - * tests/version_tests/*.dsp: We should have used ACE_NO_INLINE - instead of defining __ACE_INLINE__=0 to supress the use of - inlining. Thanks to Nick Sawadsky - <Nick_Sawadsky@BrooksSoftware.com> for reporting this. - -Fri May 07 23:05:24 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-vxworks5.x.h: added ACE_LACKS_LINEBUFFERED_STREAMBUF - for Green Hills 1.8.9. - - * ace/README: added ACE_LACKS_LINEBUFFERED_STREAMBUF. - -Fri May 7 19:39:43 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Message_Block_T.i: Found little buf in - ACE_Locked_Data_Block<> default constructor. - -Fri May 07 16:54:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.{cpp, h, i}: Reorganized and optimized internal - methods, added read and write guards, implemented all public - find and unbind methods for the interface comparable to - ACE_Hash_Map_Manager_Ex. - -Fri May 07 14:40:12 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_osf1_cxx.GNU: added, to support - uniform use of cxx across different OS versions. - - * ace/config-osf1-4.0.h: conditionally #define DIGITAL_UNIX, to - support definition in CFLAGS. - - * ace/config-osf1.h: added, to support easy migration to DU 5.0. - -Fri May 07 11:00:52 1999 David L. Levine <levine@cs.wustl.edu> - - * ACE-INSTALL.html: OK on RedHat 6.0 Linux. - -Fri May 07 08:58:47 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.{h,i}: added some QNX support. In OS.i, this mostly - consisted of adding a !defined (ACE_LACKS_MUTEXATTR_PSHARED) - check to _POSIX_THREAD_PROCESS_SHARED. Thanks to Alain Magloire - <alain@qnx.com> for providing this support. - -Thu May 6 21:48:26 1999 Ossama Othman <othman@cs.wustl.edu> - - * acconfig.h: - * configure.in: - * ace/OS.h: - * ace/README: - * ace/config-hpux11.h: - * ace/config-irix6.x-g++.h: - * ace/config-irix6.x-kcc.h: - * ace/config-irix6.x-sgic++.h: - * ace/config-linux-lxpthreads.h: - * ace/config-osf1-4.0.h: - * ace/config-sco-5.0.0-mit-pthread.h: - * ace/config-sunos5.5.h: Removed all references to - ACE_HAS_PTHREAD_SIGMASK. This macro has been deprecated. - -Thu May 6 20:12:57 1999 Ossama Othman <othman@cs.wustl.edu> - - * acconfig.h: - * configure.in: Added tests/support for ACE_LACKS_PTHREAD_SIGMASK, - ACE_LACKS_TCP_NODELAY and ACE_LACKS_NULL_PTHREAD_STATUS. Thanks - to David for providing a test for ACE_LACKS_NULL_PTHREAD_STATUS. - -Thu May 06 14:26:54 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/README,ace/Pipe.cpp (open): added support for - ACE_LACKS_TCP_NODELAY. Thanks to Alain Magloire <alain@qnx.com> - for providing the patch. - -Thu May 6 14:23:14 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/Synch_T.cpp: Removed ACE_ALLOC_HOOK_DEFINE macros for the - Guard<> objects, they don't need them (because they are - allocated on the stack) and the HOOK_DECLARE macros where not - used in the header file. - - * bin/count_lines: Several new file types, fixed computations when - multiple modules are used. - -Thu May 06 12:28:54 1999 David L. Levine <levine@cs.wustl.edu> - - * docs/ACE-FMM.html: replaced a > with > and added RCS Id - string. - - * ACE-INSTALL.html: added link to docs/ACE-FMM.html. - -Thu May 06 12:23:51 1999 Steve Huston <shuston@riverace.com> - - * ace/OS.i (recvv, sendv): Added code to make these functions work - on Win32 w/ Winsock 1.1. - -Wed May 05 15:10:37 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: added ACE_U_ONE_SECOND_IN_*SECS. They're typed and - unsigned. The old ACE_ONE_SECOND_IN_*SECS macros are retained - for backward compatibility. - -Wed May 5 13:58:25 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/OS.i: Don't use poll() for the implementation of - ACE_OS::sleep(), it has bad resolution and it gives wrong - results on Linux. - - * tests/High_Res_Timer_Test.cpp: Added new option to run several - iterations of each interval. - -Wed May 05 13:14:52 1999 Steve Huston <shuston@riverace.com> - - * ace/ACE.cpp: Removed DllMain - Win32 does not have the gusto to - handle the initialization/finalization in this way. It causes - DLL dependency problems which result in early detachment of - winsock, and possible network data loss in some cases. - - * ace/OS.cpp, Object_Manager.cpp: Clarified some comments - regarding multiple ACE_OS_Object_Manager and ACE_Object_Manager - objects in existence and how they're handled. - - * tests/ACE_Init_Test.cpp: Always does ACE::init and ACE::fini - now. This is one example of how to handle a non-console program - in Win32. - -Wed May 05 13:07:49 1999 David L. Levine <levine@cs.wustl.edu> - - * bin/Makefile: replaced /bin/rm with $(RM) and removed clean - pseudotarget. - -Tue May 04 20:38:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * tests/RB_Tree_Test.cpp: Added missing explicit template - instantiations for ACE_RB_Tree_Iterator_Base. Thanks to Carlos - O'Ryan for reporting this. - - * include/makeinclude/platform_psos_diab_ppc.GNU: added -g flag - back into debug flags since it can easily be deactivated via - make debug=0. - -1999-05-04 Seth Widoff <swidoff@cs.cmu.edu> - - * ace/Containers_T.cpp (ACE_Array_Base<T>::max_size): When - resizing the array to be larger, the max_size method was copying - in the contents of the older, smaller array, but not - initializing the new elements. - -Tue May 4 17:37:30 1999 Kirthika Parameswaran <kirthika@cs.wustl.edu> - - * examples/Web_Crawler/Options.cpp (parse_args): Change the - default options from "www.cs.wustl.edu" to "tao.cs.wustl.edu". - -Tue May 04 16:24:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.{cpp, h, i}: Added deprecation comments to methods - that are going to be replaced by the new Hash_Map_Manager - compliant interface. Factored out iterator base class, added - reverse iterator. Interim checkin since it all compiles and - RB_Tree_Test runs clean. - -Tue May 04 15:56:41 1999 Steve Huston <shuston@riverace.com> - - * ace/ACE.cpp: Re-enabled DllMain (see Mon May 3 entry from Chris - Gill). - - * ace/OS.cpp, Object_Manager.cpp: In ACE_OS_Object_Manager and - ACE_Object_Manager, differentiate between The singleton instance - and other instances when allocating and freeing per-process - information. - - * tests/run_tests.bat: Added ACE_Init_Test. - -Mon May 3 18:04:32 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * examples/Reactor/Proactor/test_aiocb_ace.cpp: Added this - file. If this test is sucessful, ACE_POSIX_AIOCB_PROACTOR could - be defined for a platform. - -Mon May 03 09:35:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/RB_Tree.cpp: fixed bug in ACE_RB_Tree::RB_delete_fixup in - which a null pointer was dereferenced while trying to determine - the color of the node that was being pointed to. Per the RB - Tree discussion in Cormen Lieserson and Rivest's "Introduction - to Algorithms", after which this implementation is modeled, a - nil node is implicity treated as having been colored BLACK. - Thanks to Long Hoang <LHoang@hwdcsaws.cahwnet.gov> for reporting - the problem and providing a test program that showed the bug. - -Mon May 03 15:53:15 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_vxworks5.x_g++.GNU, - tests/run_tests.vxworks,ACE-INSTALL.html: added support for - shared lib builds. Thanks to Ramiro Penataro Blanco - <penataro@ll.iac.es> and James D. Rucker - <jdrucker@huey.jpl.nasa.gov> for figuring this out, and - providing detailed instructions for others to follow. - -Mon May 03 15:20:00 1999 Kirthika Parameswaran <kirthika@bambuca.cs.wustl.edu> - - * examples/Web_Crawler/main.cpp: - * examples/Web_Crawler/Web_Crawler.{h,cpp}: - * examples/Web_Crawler/Cached_Connect_Strategy_T.{h,i,cpp}: - * examples/Web_Crawler/Caching_Strategies_T.{h,i,cpp}: - * examples/Web_Crawler/Command_Processor.{h,cpp}: - * examples/Web_Crawler/HTTP_URL.{h,cpp}: - * examples/Web_Crawler/Iterators.{h,cpp}: - * examples/Web_Crawler/Mem_Map_Stream.{h,cpp}: - * examples/Web_Crawler/Optimal_Cache_Map_Manager_T.{h,i,cpp}: - * examples/Web_Crawler/Options.{h,cpp}: - * examples/Web_Crawler/URL.{h,cpp}: - * examples/Web_Crawler/URL_Addr.{h,cpp}: - * examples/Web_Crawler/URL_Status.{h,cpp}: - * examples/Web_Crawler/URL_visitor.{h,cpp}: - * examples/Web_Crawler/URL_Visitor_Factory.{h,cpp}: - * examples/Web_Crawler/Makefile - * examples/Web_Crawler/README - - This example is a HTTP client which crawls over the link provided - and checks whether the links recusively are valid or not. - This example was used to test teh auto-purging feature where the - connections are removed from the connection cache when the process - runs out of file descriptors. - -Mon May 03 14:46:28 1999 Steve Huston <shuston@riverace.com> - - * tests/ACE_Init_Test.dsp: Fixed to build on all configurations (for - Intel anyway). - -Mon May 03 10:30:37 1999 David L. Levine <levine@cs.wustl.edu> - - * ACE version 4.6.38 released. - -Mon May 3 09:56:06 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/INET_Addr.cpp (get_host_name): Make sure to check if - get_host_name() fails and do something useful in this case. - Thanks to Frank O'Dwyer <fod@brd.ie> for reporting this. - -Mon May 03 09:35:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/ACE.cpp: disabled BOOL APIENTRY DllMain(HANDLE hModule, - DWORD ul_reason_for_call, LPVOID lpReserved) by having it - just return TRUE. This function is used only on NT. This - is a temporary workaround for the problem of tao_idl - crashing on NT, and will be replaced ASAP with a more - complete fix to Object Manager initialization/cleanup at - DLL load/unload time on NT. Thanks to David Levine for - suggesting this fix. - -Mon May 03 09:27:54 1999 David L. Levine <levine@cs.wustl.edu> - - * docs/ACE-guidelines.html: added guideline to put - spaces around comment delimiters. Thanks to Chris - Gill for suggesting this. - -Sun May 02 14:36:43 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: Added ghs back to the list of compilers for which - ACE_NOTREACHED disappears. It was causing warnings for - many ACE and TAO files. - -Sun May 02 08:21:13 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i (pthread_sigmask): added ACE_UNUSED_ARGs to - NOTSUP case. - - * ace/OS.i (pthread_sigmask),ace/config-chorus.h,ace/README: - added ACE_LACKS_PTHREAD_SIGMASK. - -Sat May 01 17:33:04 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.cpp,Object_Manager.cpp: reverted this change because - it causes a seg fault on NT: - - Thu Apr 29 13:41:52 1999 Steve Huston <shuston@riverace.com> - - * ace/{OS.cpp Object_Manager.cpp}: Fixed leaking ACE_Object_Manager - when config has #define ACE_HAS_NONSTATIC_OBJECT_MANAGER 0. - -Sat May 01 12:06:57 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ACE.cpp: - * ace/config-win32.h: Reverted my previous change. I didn't - notice David had already fixed it. Thanks David. - -Sat May 01 01:48:29 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/ACE.cpp (DllMain): Only use DllMain to control - Object_Manager's lifecycle when - ACE_USES_DLL_TO_MANAGE_THE_LIFECYCLE_OF_OBJECT_MANAGER is - defined and we are building ACE as a DLL. - - * ace/config-win32.h: Defining - ACE_USES_DLL_TO_MANAGE_THE_LIFECYCLE_OF_OBJECT_MANAGER =1, when - compiling DLL version of ACE will cause then ACE::init/fini be - called automatically for you in DllMain when ACE gets linked - in/unlinked. - - Defining ACE_USES_DLL_TO_MANAGE_THE_LIFECYCLE_OF_OBJECT_MANAGER - =1 automatically define ACE_HAS_NONSTATIC_OBJECT_MANAGER to 1 - and define ACE_DOESNT_INSTANTIATE_NONSTATIC_OBJECT_MANAGER. - -Fri Apr 30 22:53:01 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.h: put Object_Manager back on the stack of main, - instead of using ACE::init () and ACE::fini (). It - should be functionally equivalent, and then some: if - the user's main () returns, the Object_Manager will get - properly destroyed if it's on the stack. And, it's better - for subset support if we avoid class ACE calls from OS.h. - - * include/makeinclude/rules.lib.GNU: cleaned up a bit. - -Fri Apr 30 17:42:11 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * ace/OS.i (pthread_sigmask): - * ace/OS.h (ACE_OS): Added wrapper for <pthread_sigmask> to get - the Proactor code compile on LynxOS. - - * ace/config-lynxos.h (ACE_HAS_AIO_CALLS): - Enabled ACE_HAS_AIO_CALLS for LynxOS. Also enabled - ACE_HAS_AIOCB_PROACTOR since LynxOS 3.0.0 has not got - <pthread_sigmask> yet, so signal proactor cannot work correctly. - - * ace/POSIX_Proactor.cpp: - * examples/Reactor/Proactor/test_aiosig_ace.cpp - (setup_signal_delivery): Added ACE_OS:: scope to pthread_sigmask - so that it works (fails!!) correctly on LynxOS. - - * examples/Reactor/Proactor/post_completions.cpp: - * examples/Reactor/Proactor/test_timeout.cpp: Added 'main' - function for platforms where only AIOCB_Proactor is enabled so - that the test prints out a message. - - * examples/Reactor/Proactor/Makefile: - -Fri Apr 30 14:44:12 1999 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ACE-INSTALL.html: Added a note about manually cleaning out the - ACE_wrappers directories on Win32, which could be required for a - true start from scratch build. Thanks to David Hooker - <DHooker@uniview.net> for this insight. - -Fri Apr 30 14:06:41 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * COPYING: Updated the COPYING file to reflect the use new of - ACE(TM) and TAO(TM). - -Fri Apr 30 13:23:22 1999 Steve Huston <shuston@riverace.com> - - * docs/tutorials/017/Barrier_i.cpp: - * docs/tutorials/018/token.cpp: Added explicit template - instantiations. - -Fri Apr 30 12:11:36 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/OS.i (gethostbyname2): - - Added an IPv4-only implementation. If ACE_HAS_IP6 is not defined, - the IPv4-only implementation will be used. - -Fri Apr 30 12:11:50 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * tests/Basic_Types_Test.cpp: Added checks for sizeof(long) == - sizeof(void*) and sizeof(long) >= sizeof (a_function_pointer). - Certain code in ACE depends on the assumptions. - -Fri Apr 30 11:24:26 1999 Steve Huston <shuston@riverace.com> - - * tests/ACE_Init_Test.cpp: Added compile-time change to adapt to - build w/ static ACE lib and ACE_HAS_NONSTATIC_OBJECT_MANAGER. - * tests/version_tests/ACE_Init_Test.dsp: Changes to build clean on - all configurations. - -Fri Apr 30 02:06:53 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Parse_Node.cpp (symbol): According to the new ANSI C++ - specification, casting a void* pointer to a function pointer is - not allowed. However, casting a void* pointer to an integer - type that is large enough to hold the pointer value is legal. I - chose to cast the return value to long since it should be large - enough to hold the void* pointer's value on most platforms. I am - not sure if casting a long value to a function pointer is legal - or not (can't find a good explanation in spec) but it prevent - SunC++ 5.0 compilers from complaining about illegal pointer - conversion. - - * ace/OS.h: Removed ACE_non_function_ptr_to_function_ptr_cast, - it is no longer needed. - -Fri Apr 30 00:36:23 1999 Ossama Othman <othman@cs.wustl.edu> - - - * ace/Get_Opt.cpp: - * ace/TLI_Acceptor.cpp: - * ace/TLI_Connector.cpp: - - Cast away const from string literals using ACE_const_cast to make - Sun C++ 5.0 happier. - -Thu Apr 29 18:32:43 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/OS.h: Removed ghs from the list of compilers for which - ACE_NOTREACHED disappears. It was causing warnings for - TAO/orbsvcs/orbsvcs/Naming/Persistent_Context_Index_T.cpp. - -Thu Apr 29 19:48:25 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/OS.{h,i} (ACE_OS): - - More IPv6 support. Added support for the new IPv4/IPv6 - inet_ntop() and inet_pton() calls. If ACE_HAS_IP6 is not - defined, these calls will use IPv4 implementations (see - Stevens' book "UNIX Network Programming - Volume I") and - return an error if a family other than AF_INET is passed to - them. - - * ace/OS.h: - - Define INET_ADDRSTRLEN and INET6_ADDRSTRLEN if they don't - get defined after including <netinet/in.h>. - -Thu Apr 29 18:32:43 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/README: - * ace/OS.h (ACE_OS): - * ace/OS.i (gethostbyname2): - - Added support for the gethostbyname2() system call. It is only - enabled if ACE_HAS_IP6 is defined. When I have time, I'll add - more IPv6 support. - -Thu Apr 29 15:30:04 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/INET_Addr.cpp (string_to_addr): Added support to check for - both "port numbers" and "port names". Thanks to Jerry D. De - Master <jdemaste@ipdinc.com> for suggesting this. - - * ace/Select_Reactor_T.h, - ace/Reactor.h, - ace/Select_Reactor_T.h, - ace/TP_Reactor.h (ACE_TP_Reactor): Updated the documentation to - clarify the purpose of the <restart> flag. Thanks to Susan - Liebeskind <shl@cc.gatech.edu> for reporting this. - -Thu Apr 29 13:41:52 1999 Steve Huston <shuston@riverace.com> - - * ace/{OS.cpp Object_Manager.cpp}: Fixed leaking ACE_Object_Manager - when config has #define ACE_HAS_NONSTATIC_OBJECT_MANAGER 0. Now - both ACE_OS_Object_Manager and ACE_Object_Manager remember if they - were dynamically allocated whether the above #define is used or not - and correctly clean up and track the singleton properly. - Thanks to David and Irfan for identifying this and helping with the - solution. - -Wed Apr 28 20:21:12 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Synch_T.h (class ACE_Anti_*_Guard): I had accidentally - checked in the anti guard classes. These are not needed since - we decided to stick with the reverse lock class rather than the - anti guard classes. - -Wed Apr 28 18:02:36 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.h: Added a new macro called - ACE_non_function_ptr_to_function_ptr_cast to get around some - compiler constrain. In face, ANSI C++ standard prevents casting - from a non-function pointer (e.g., void *) to a function - pointer. Currently, only egcs check for this condition. - - * ace/Parse_Node.cpp (symbol): Changed to use the new - ACE_non_function_ptr_to_function_ptr_cast. - -Wed Apr 28 16:53:16 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * examples/Reactor/Proactor/Aio_Platform_Test_C.cpp: Added this - file, a simple C version of the - $ACE_ROOT/tests/Aio_Platforms_Test.cpp. Useful for submitting bug - reports etc. - - * ace/Proactor.cpp: - * ace/POSIX_Asynch_IO.cpp: Fixed old g++ warnings - -Wed Apr 28 15:39:14 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Parse_Node.cpp (symbol): Use reinterpret_cast to cast - dlsym's return value. - -Wed Apr 28 12:40:26 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Synch_T.{i,cpp} (~ACE_Reverse_Lock): uninlined destructor. - See docs/ACE-guidelines.html for an explanation of why a - class should have at least one non-inline, non-pure virtual - function if it has any virtual functions. egcs 1.1.1b - needed this one. - -Wed Apr 28 09:02:26 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/config-sunos5.5.h: It appears that SunC++ 5.0 wants - ACE_HAS_THR_C_DEST to be defined. - - * ace/SString: Clarified the comments a bit to avoid confusion. - Also, moved the definition of ACE_SString to AFTER ACE_CString - and ACE_WString since they are more general. Thanks to Kevin - Lyda <kevin.lyda@trintech.com> for reporting this. - -Tue Apr 27 20:59:52 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * netsvcs/clients/Naming: Moved the README file into the Client/ - directory since that's what it described. - - * netsvcs/clients/Naming/Client/Client_Test.cpp: Reformatted some - code to conform to ACE programming guidelines. - - * ace/Naming_Context.cpp (parse_args): Oops, fixed a braino - related to strcmp(). Thanks to Valery Arkhangorodsky - <avalery@balisoft.com> for reporting this. - -Tue Apr 27 18:00:31 1999 Steve Huston <shuston@riverace.com> - - * ace/OS.cpp (ACE_OS_CString, ACE_OS_WString): Fixed one-off error - in allocating new string's memory. - -Tue Apr 27 17:36:28 1999 Ossama Othman <othman@cs.wustl.edu> - - * m4/threads.m4: - - Restructured thread detection test so that it properly detects - threads on platforms that provide PTHREADS backward compatibility - macros, such as Digital Unix 4.0 and 5.0. - -Tue Apr 27 18:56:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Containers_T.{cpp, h} - tests/OrdMultiSet_Test.cpp: added a workaround for a problem with - implicit template instantiation and a do-nothing destructor - definition. Thanks to Sarmeesha Reddy <sreddy@tr.comm.mot.com> and - Steven Tine <stevent@tr.comm.mot.com> for reporting this. - - * include/makeinclude/platform_psos_diab_ppc.GNU: changes to use - the board support package linker file. Thanks to Sarmeesha - Reddy <sreddy@tr.comm.mot.com> and Steven Tine - <stevent@tr.comm.mot.com> for contributing this change. - -Tue Apr 27 16:30:57 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/UNIX_Addr.h (ACE_UNIX_Addr): - - Corrected typographical error in comment. - -Tue Apr 27 12:57:06 1999 Ossama Othman <othman@cs.wustl.edu> - - * configure.in: - - Removed `tests/log/Makefile' from the list of files to - generate (i.e. removed from AC_OUTPUT macro call). - - * tests/Makefile.am: - - Corrected a typo. I neglected to add "_SOURCES" to the end - of the Collection_Test source variable variable. Also added - more missing tests to the list of test programs. Commented out - Cache_Manager_Test. It appears to need a header that doesn't - exist. - - * ace/Makefile.am (pkginclude_HEADERS): - - Removed `Timer_Queue.i' from the list of headers. - - * bin/autogen: - * bin/bootstrap: - - Renamed the `autogen' script to `bootstrap'. - -Tue Apr 27 11:54:53 1999 Ossama Othman <othman@cs.wustl.edu> - - * configure.in: - - Temorarily removed my experimental C++ libtool support so that - I can get a semi-working test distribution out on the net. - - * ltconfig: - * ltmain.sh: - - Reverted to more stable version, i.e. w/o my C++ support. - - * m4/threads.m4: - - Added KAI C++ thread flag check. - -Tue Apr 27 11:45:40 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/Collection_Test.cpp (main): Added some additional array - tests and fixed template instantiation errors. Thanks to David - Levine for pointing them out. - - * ace/OS.h (ACE_DES_FREE): Changed all the ACE_DES* macros to - check for zero pointers. Thanks to Torsten Kuepper - <kuepper2@uni-wuppertal.de> for pointing this bug out. - -Tue Apr 27 11:08:35 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Basic_Types.h: Forced KAI compiler to use ACE_U_LongLong if - ACE_LACKS_LONGLONG_T is defined. - - * ace/DLL.cpp: - * ace/Service_Config.i: - * ace/Service_Object.i: - * ace/INET_Addr.cpp: Some CE related changes. - -Tue Apr 27 00:03:16 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/Collection_Test.cpp (main): Added test code for ACE_Array. - -Mon Apr 26 23:55:43 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Synch_T.h (class ACE_Reverse_Lock): Removed constructor that - created the lock. There were two reasons for this: (a) it was - unrealistic that the reverse lock would own and create the - regular lock and (b) this constructor made it impossible that - ACE_Reverse_Lock can be used with ACE_Lock (the abstract - class). Thanks to Carlos for pointing this out. - -Mon Apr 26 21:43:23 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/Makefile.am (pkginclude_HEADERS, libACE_IPC_la_SOURCES): - - Added ATM_Addr sources to these variables. - - * ace/Makefile.am (EXTRA_libACE_la_SOURCES): - - Accidentally removed this variable. It is needed for - conditionally built sources, such as `gethrtime.cpp'. - - * tests/Makefile.am: - - Added newly added tests to the list of test programs to be built. - Changed "noinst_PROGRAMS" variable to "check_PROGRAMS" so that - tests only get built when a `make check' is issued by the user. - -Mon Apr 26 19:29:25 1999 Ossama Othman <othman@cs.wustl.edu> - - * tests/Reverse_Lock_Test.cpp (main): - - ACE_START_TEST macro had "Thread_Mutex_Test" instead of - "Reverse_Lock_Test." - -Mon Apr 26 17:01:22 1999 Steve Huston <shuston@riverace.com> - - * ace/ACE.cpp (DllMain): Re-enabled this for use on Win32. - - * ace/OS.h (main): On non-pSoS, use ACE::init, ACE::fini rather than - creating object manager instances on the stack. Allows a single - instance regardless of program type; especially helpful on Win32. - - * ace/OS.cpp (ACE_OS_Object_Manager ctor): - * ace/Object_Manager.cpp (ACE_Object_Manager ctor): - Don't set instance_ if it was already set. So if two - ACE_Object_Manager instances end up getting created, the first one - stays The Instance, and the second gets used privately by its - creator. Previously, the original was forgotten. Thanks to David - Levine for working out this and the above details with me to get the - Win32 crowd in business without platform-specific hacks. - -Mon Apr 26 16:43:32 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Asynch_IO_Impl.h (ACE_Asynch_[Read|Write]_File_Impl): - Declared the read/write pure virtual methods of - ACE_Asynch_[Read|Write]_Stream_Impl again in these class to - avoid KAI compiler's overwriting pure virtual function - warnings. - -Mon Apr 26 10:54:23 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * ace/High_Res_Timer.h: - Moved the static method hrtime_to_tv to the public section. It - is useful outside this class. - - * ace/Synch_T.cpp: - The default constructor for ACE_Reverse_Lock should not allocate - the locking mechanism, otherwise it cannot work with ACE_Lock. - -Mon Apr 26 07:30:51 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/OS.i (fstat): Changed to use CE's implementation and added - some features that're supported on other Win32 platforms. The - original implementation for Win32 closed down the file handle - accidentally which is not a "Good Thing[TM]". Thanks to Dominic - Williams <dom@connected-place.co.uk> for pointing this out. - -Sun Apr 25 21:35:37 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/ACE.{h,cpp},OS.{h,cpp},Object_Manager.{h,cpp} (init,fini): - moved init_fini_count_ from ACE Object_Managers to ACE, so that - it is only used via ACE::init () and ACE::fini (). Thanks - to Steve Huston for pointing out that the old way didin't work - if an Object_Manager was created prior to starting main (), - and there were ACE::init ()/ACE::fini () calls in main (). - -Sun Apr 25 16:59:21 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * include/makeinclude/platform_sunos5_kcc.GNU: Fixed incorrect - compiler flags. - - * ace/config-sunos5.7.h: Added definition of uint64_t on KAI. - - * ace/config-kcc-common.h: Changed to use non-static object - manager. - Defined ACE_SIZEOF_LONGLONG. KAI compiler has type long long. - - * ace/OS.h: ACE_NOTREACHED should expand to nothing on KAI - compiler. - - * ace/CDR_Stream.h (ACE_CDR): KAI compiler doesn't have - longlong_t. - - * ace/Basic_Types.h: KAI compiler should define ACE_UINT64 as - unsigned long long. - -Sun Apr 25 14:52:18 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * COPYING, - TAO/COPYING: Updated the ACE+TAO COPYING information to - emphasize the relationship to open source. Thanks to Malcolm - Spence <spence_m@ociweb.com> for suggesting the wording. - -Sat Apr 24 23:03:31 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * ACE_wrappers/ace/POSIX_Asynch_IO.cpp - * ACE_wrappers/ace/POSIX_Asynch_IO.h - * ACE_wrappers/ace/POSIX_Proactor.cpp - * ACE_wrappers/ace/POSIX_Proactor.h - * ACE_wrappers/ace/Proactor.cpp - * ACE_wrappers/ace/Proactor.h - * ACE_wrappers/ace/Proactor_Impl.h - - * ACE_wrappers/examples/Reactor/Proactor/Makefile - * ACE_wrappers/examples/Reactor/Proactor/README - * ACE_wrappers/examples/Reactor/Proactor/test_aiocb.cpp - * ACE_wrappers/examples/Reactor/Proactor/test_aiosig.cpp - * ACE_wrappers/examples/Reactor/Proactor/test_aiosig_ace.cpp - * ACE_wrappers/examples/Reactor/Proactor/test_proactor.cpp - * ACE_wrappers/examples/Reactor/Proactor/test_timeout.cpp - - - Completed Multithreading of POSIX_SIG_Proactor. - - - Introduced another constructor for ACE_POSIX_SIG_Proactor for taking - signal mask to be used with the Proactor. - - - Thanks to Dave Butenhof <butenhof@zko.dec.com> for helping a lot to - understand the various things in the POSIX4 standard. - - - Thanks to Dave suggestion of keeping null_handler for the sigaction - to real-time signals. With this and a couple of other correct POSIX - things SIG proactor is now working with in Solaris 2.7. - - - Lynx OS doesnt support <pthread_sigmask>, so it couldnt be - multithreaded. Enabled AIOCB_Proactor for this platform. - - - Added an example to make use of the real-time signal numbers for the - asynchronous I/O calls. - - * examples/Reactor/Proactor/test_posix_sig_proactor.cpp: This - program demonstrates how to post fake completions to The - Proactor. It also shows the how to specify the particular - real-time signals to post completions. - - * ace/config-lynxos.h: Enabled ACE_POSIX_AIOCB_PROACTOR for this - platform, since <pthread_sigmask> was not available. Without - <pthread_sigmask>, SIG_Proactor can not work correctly. - - * ace/config-linux-lxpthreads.h: Removed ACE_HAS_AIO_CALLS. Even - simple aio calls test is hanging. - -Sat Apr 24 13:09:27 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/Containers_T.h: Added allocator support to ACE_Array. - Thakns to Paul Francis <pfrancis@dkl.com> for contributing this. - - * ace/Future.h (class ACE_Future): Make sure to mention that - get()'s ACE_Time_Value must be in absolute, not relative, time. - Thanks to Ian MacDonald <ianmacd@bg.com> for reporting this. - -Sat Apr 24 07:24:51 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Reverse_Lock_Test.cpp: removed ACE_Guard<ACE_SYNCH_MUTEX> - instantiation because it's in ace/Synch.cpp. - - * ace/config-vxworks5.x.h: added ACE_THR_PRI_OTHER_DEF - #define, for use by MT_Cubit. - - * ace/Object_Manager.cpp, OS.cpp (fini): unconditionally set - instance_ to 0 at the end. This allows init () to - work properly if called after fini (), such as for - repeated test on VxWorks. Thanks to Todd Mullanix - <Todd.Mullanix@sylantro.com> for reporting this. - -Fri Apr 23 17:53:24 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_vxworks5.x_g++.GNU: - for PPC860, changed -m=860 to -mcpu=860. Thanks to - Todd Mullanix <Todd.Mullanix@sylantro.com> for reporting this. - -Fri Apr 23 15:48:31 1999 James Hu <jxh@entera.com> - - * ace/OS.h: - * ace/OS.i: - * ace/README: - * ace/config-sunos5.5.h: added support for the memchr - library call, and created ACE_HAS_MEMCHR macro. - -Fri Apr 23 10:48:31 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/New_Fail_Test.cpp (main): added uses of the - two static functions when they're not otherwise used, - to prevent g++ warnings. - -Fri Apr 23 09:11:24 1999 Carlos O'Ryan <coryan@cs.wustl.edu> - - * tests/Reverse_Lock_Test.cpp: - Fixed template instantiation problems. - -Thu Apr 22 21:39:44 1999 Ossama Othman <othman@cs.wustl.edu> - - * tests/Makefile.am: - - Added support to create the `tests/log' directory after - the tests have been built. - - * tests/log/Makefile.am - - The above change deprecates this Makefile.am file. Thanks to - Nanbor for motivating this change. - -Thu Apr 22 20:50:30 1999 Jeff Parsons <parsons@cs.wustl.edu> - - * ace/CDR_Stream.i: Modified reset() to reset all the message - blocks, instead of just the first one as it had previously done. - Thanks to Jerry D. De Master <jdemaste@ipdinc.com> for pointing - out the problem and suggesting a fix. - -Thu Apr 22 19:40:52 1999 Steve Huston <shuston@riverace.com> - - * ace/OS.h: Adjust exception thrown at out-of-memory for HP-UX, aC++. - - * ace/config-hpux-10.x-hpc++.h: Add ACE_NEW_THROWS_EXCEPTIONS. - -Thu Apr 22 19:30:01 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Synch_T.h (class ACE_Reverse_Lock): Added an interesting - adapter class that changes a lock into a reverse lock, i.e., - acquire() on this class calls release() on the lock, and - release() on this class calls acquire() on the lock. - - One motivation for this class is when we temporarily want to - release a lock (which we have already acquired) but then - reaquire it soon after. An alternative design would be to add a - Anti_Guard or Reverse_Guard class which would release() on - construction and acquire() destruction. However, there are - *many* varieties of the Guard class and this design choice would - lead to at least 6 new classes. One new ACE_Reverse_Lock class - seemed more reasonable. - - * tests/Reverse_Lock_Test.cpp: Test for the new ACE_Reverse_Lock - class. - - * tests/Env_Value_Test.cpp (main): Fixed UNICODE string - concatenation problem. - - * tests/version_tests: Added a bunch of missing project files. - - * ace/ace_lib.dsp: Added missing files. - -Thu Apr 22 19:10:52 1999 Steve Huston <shuston@riverace.com> - - * ace/ATM_Addr.cpp: Put in class's consts for non-FORE platforms. - -Thu Apr 22 17:05:36 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * bin/auto_compile_win32.pl: Added version_tests. - - * include/makeinclude/platform_sunos5_kcc.GNU: Makefile macros for - KAI C++ compiler. - - * ace/config-sunos5.5.h: Added KAI C++ compiler support. - - * docs/exceptions.html: Explicitly state that ACE try macros will - replace TAO try macros and advise users to take action if they - are still using TAO try macros. Thanks to Don Busch - <busch_d@ociweb.com> for pointing this out. - -Thu Apr 22 15:07:10 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/platform_vxworks5.x_g++.GNU: - added PPC860 support. Thanks to Todd Mullanix - <Todd.Mullanix@sylantro.com> figuring out how to - support it, with help from Thomas Mehrkam <Thomas_Mehrkam@i-o.com> - and Hai Vu <Hai_Vu@Mitel.Com>. - -Thu Apr 22 13:26:26 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/XtReactor.cpp (remove_handler_i): In the registration phase, - we registered first with ACE_Select_Reactor and then with X. - Now we are now doing things in reverse order. Thanks to JM - Strauss <jms97@club-internet.fr> for contributing these changes. - -Thu Apr 22 13:23:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * include/makeinclude/platform_psos_diab_ppc.GNU: commented out -g - switch which was triggering a bug in the Diab 4.2b compiler when - compiling in debug information. This is supposed to be fixed in - the 4.3a release of the Diab compiler. The ACE tests all compile - now for pSOS/PPC. Thanks to Sarmeesha Reddy <sreddy@tr.comm.mot.com> - for reporting the bug to Diab and obtaining the workaround. - -Thu Apr 22 10:46:17 1999 Steve Huston <shuston@riverace.com> - - * apps/JAWS/server: Added $(ACELIB) to LDLIBS to build libJAWS on AIX - with xlC. - - * tests/New_Fail_Test.cpp: Added __GNUG__ to the platforms which do not - actually perform the test; added some more explanation and a better - log message for when it doesn't actually run. - -Thu Apr 22 10:28:39 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-minimal.h: removed suppression of ACE_TSS_EMULATION - and enable of ACE_USE_THREAD_MANAGER_ADAPTER. They're no - longer necessary with the clean OS.cpp. - -Thu Apr 22 08:52:48 1999 Steve Huston <shuston@riverace.com> - - * tests/New_Fail_Test.cpp: Fixed to compile clean without exceptions. - - * tests/run_tests.{sh bat}: Added New_Fail_Test. - -Wed Apr 21 16:41:08 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Reactor_Performance_Test.cpp (main): added 10 second - time limit to run_event_loop () call. That prevents the - test from hanging forever if some of the connections - are lost. - - * tests/Priority_Reactor_Test.cpp: default to 5 children - instead of 10 on LynxOS. Beyond 8, the test seems to - take forever on LynxOS 3.0.0. - - * tests/run_tests.sh: enabled Reactor_Performance_Test, - Priority_Reactor_Test, Time_Service_Test, and Tokens_Test - on LynxOS. - -Wed Apr 21 16:35:21 1999 Steve Huston <shuston@riverace.com> - - * tests/New_Fail_Test.cpp: Test to be sure that the ACE_NEW[_RETURN] - stuff works correctly when heap is exhausted. - * tests/Makefile: Added New_Fail_Test. - * tests/tests.dsw, New_Fail_Test.dsp: Add new test to MSVC. - -Wed Apr 21 15:35:44 1999 Steve Huston <shuston@riverace.com> - - * ace/ACE.cpp: On Win32, with ACE_HAS_NONSTATIC_OBJECT_MANAGER, use - DllMain to do ACE::init() and ACE::fini(). - - * tests/ACE_Init_Test.dsp, - * tests/ACE_Init_Test.{cpp h}, - * tests/ACE_Init_TestDlg.{cpp h}, - * tests/ACE_Init_Test.(ico rc rc2 res}, - * tests/ACE_Init_Test_(Resource StdAfx}.h: New test to be sure that - ACE DLL is initialized properly from an MFC app. - - * tests/tests.dsw: Added ACE_Init_Test.dsp project. - -Wed Apr 21 14:59:04 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.cpp (ACE_OS_Object_Manager::fini): on LynxOS only, - disabled ACE_{recursive/thread}_mutex_destroy () calls, because - they failed on LynxOS in forked children. The prevents - the failure message from being printed, with no apparent - ill effect. - -Wed Apr 21 10:06:11 1999 Steve Huston <shuston@riverace.com> - - * include/makeinclude/platform_sunos5_sunc++.GNU: Moved -lthread from - the all-Orbix section to the MT-Orbix section for LIBS. - -Wed Apr 21 09:25:16 1999 David L. Levine <levine@cs.wustl.edu> - - * tests/Upgradable_RW_Test.cpp (main): replaced ACE_SCHED_FIFO - with THR_SCHED_FIFO in activate call. Thanks to Carlos for - finding this subtle error. - -Wed Apr 21 03:34:18 1999 Douglas C. Schmidt <schmidt@cs.wustl.edu> - - * ACE version 4.6.37 released. - -Tue Apr 20 22:55:18 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.i (thr_join): ifdef ACE_LACKS_PTHREAD_NULL_STATUS, - provide the address of a temporary variable for the status - argument of ::pthread_join (), if the caller passed 0. - - * ace/config-lynxos.h,README: added ACE_LACKS_PTHREAD_NULL_STATUS. - LynxOS 3.0.0 fails ::pthread_join () calls with EFAULT - if the second (status) argument is 0. - -Tue Apr 20 21:53:32 1999 Joe Hoffert <joeh@cs.wustl.edu> - - * examples/IPC_SAP/TLI_SAP/Makefile: Had inadvertently checked in - local changes. Now undoing these changes. - -Tue Apr 20 18:25:45 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/{SOCK,TLI,LSOCK,Asynch}_Acceptor.h: Replaced the use of the - hard-coded backlog value of 5 with ACE_DEFAULT_BACKLOG. Thanks - to Irfan for motivating this. - - * ace/OS.h: Added a new (overriddable) macro called - ACE_DEFAULT_BACKLOG that's used to control the default number of - connections that can be accepted by an OS. The default value is - 5. - -Tue Apr 20 18:22:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/Map_T.i - ace/config-psos-diab-mips.h - ace/config-psos-diab-ppc.h - ace/config-psos-diab.h - include/makeinclude/platform_psos_diab_ppc.GNU - tests/Handle_Set_Test.cpp - tests/Process_Strategy_Test.cpp: fixed the remaining tests - for which compiler bug workarounds were available for pSOS - PPC with the Diab 4.2b compiler. - -Tue Apr 20 16:41:39 1999 Nanbor Wang <nanbor@cs.wustl.edu> - - * ace/Thread_Manager.cpp (num_tasks_in_group): Make sure we are - not searching for a null task. Thanks to Umar Syyid - <usyyid@hns.com> for pointing this out. - -Tue Apr 20 15:10:33 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/Reactor_Performance_Test.cpp: Changed the default number - of connections (<opt_nconnection>) from 20 to 5 for NT. This - seems reasonable since the default backlog is 5. Once we have - the ability to set the backlog from the ACE_Acceptor class, we - can increase this number. - -Tue Apr 20 13:55:34 1999 Joe Hoffert <joeh@cs.wustl.edu> - - * ace/OS.i: Fixed ACE_OS::t_free to free memory if pointer is not - null. - - * ace/TLI.h: Added default ACE_XTI_ATM_DEVICE. - - * ace/TLI_Connector.cpp: Disabled non-blocking connects for XTI/ATM - since FORE's drivers have problems with this. Also, fixed t_alloc - memory leaks. - - * ace/ATM_Addr.cpp: Added code formatting change. - - * examples/IPC_SAP/TLI_SAP/CPP-ATM-{server,client}.cpp: Added - the use of ACE_XTI_ATM_DEVICE - -Tue Apr 20 13:30:51 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/ace.icc: Added a configuration file for creating a shared - library with Visual Age C++. Thanks to Rene Matteau - <matteau@ca.ibm.com> for this. - -Mon Apr 19 23:50:25 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/Time_Service_Test.cpp (main): Fixed error message. - - * ace/Connector.cpp (handle_close): Removed unused argument - <mask>. - -Mon Apr 19 17:17:47 1999 Steve Huston <shuston@riverace.com> - - * docs/tutorials/tutorials.dsw: Fixed project file name for 019-021. - -Mon Apr 19 17:02:42 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Connector.cpp (handle_close): There is no need to call - cancel_timer() or remove_handler() from handle_close(). Removal - of all svc tuples from the <reactor> is done by <cleanup_AST>. - -Mon Apr 19 16:45:54 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.{h,cpp},Object_Manager.{h,cpp} (ACE_OS_Object_Manager, - ACE_Object_Manager: added static counter to match init ()/fini () - calls. Don't destroy Object_Managers until final fini () call. - Thanks to Ian MacDonald <ianmacd@bg.com> for tracking down - the problem with multiple DLLs in a process. This change - delays destruction of the ACE_Object_Manager until the final - fini () call. - -Mon Apr 19 15:24:55 1999 Douglas C. Schmidt <schmidt@tango.cs.wustl.edu> - - * ace/TLI_Stream.cpp (ACE_TLI_Stream): Make sure to set the value - of rwflag_ to 0 so that purify is happy. Thanks to Joe Hoffert - for reporting this. - -Mon Apr 19 14:35:42 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * examples/IPC_SAP/UPIPE_SAP/ex2.cpp (supplier): - * examples/IPC_SAP/UPIPE_SAP/ex3.cpp (consumer): - - Avoided annoying g++ warning about <auto_ptr> by explicitly - making a temporary. Thanks to David for pointing this out. - -Sun Apr 18 21:52:43 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/config-linux-common.h: moved ACE_LACKS_MADVISE to - non-glibc section. Thanks to JM Strauss <jms97@club-internet.fr> - for reporting that it wasn't on a Linux system with a - 2.0.35 kernel. - -Sun Apr 18 16:36:30 1999 Ossama Othman <othman@cs.wustl.edu> - - * ace/config-linux-common.h (ACE_LACKS_MADVISE): - - Re-enabled #define ACE_LACKS_MADVISE. Linux doesn't appear - to have an madvise() system call. - - * configure.in: - * tests/Makefile.am: - - Added libtool support for modules (dlopened libraries). - -Sun Apr 18 08:48:37 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/Auto_Ptr.i: reordered inline methods to avoid use before - definition. Fixed a few ACE_TRACE comments. - -Sun Apr 18 00:26:28 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/Notify_Performance_Test.cpp (main): - * tests/Reactor_Performance_Test.cpp (main): - * tests/Priority_Reactor_Test.cpp (main): - - Avoided annoying g++ warning about <auto_ptr> by explicitly - making a temporary. Thanks to David for pointing this out. - -Sat Apr 17 18:52:38 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Memory_Pool.cpp (acquire): Changed use of auto_ptr. - - * ace/Auto_Ptr.i (operator=): Changed <a> to <rhs>. Thanks to - David Levine and Jean-Marc Strauss <strauss@club-internet.fr> - for reporting this. - -Sat Apr 17 08:34:22 1999 David L. Levine <levine@cs.wustl.edu> - - * examples/Naming/Makefile,performance-tests/Misc/Makefile, - tests/Makefile,all netsvcs Makefiles: - updated for proper support of ACE subsets, with BIN2/LIB2/SHLIB2. - -Fri Apr 16 22:14:36 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * ace/Auto_Ptr: Made several changes to bring the implementation - up to the specification: - - - Made the constructors take non-const parameters. - - Made release() non-const. - - Added reset(). - - Thanks to Simeon Simeonov <simeons@allaire.com>, Rainer Blome - <blome@de.ibm.com>, Jody Hagins <jody@atdesk.com>, and John - Mulhern <John.Mulhern@lawson.com> for helping. - - * tests/Reactor_Performance_Test.cpp: Updated to work with new - auto pointer implementation. - -Fri Apr 16 19:36:35 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> - - * ace/Proactor.h (ACE_Proactor): Switched the order of TIMER_QUEUE - and delete_implementation in the constructor. - - * examples/Reactor/Proactor/test_multiple_loops.cpp: Fixed to work - on the WIN32 correctly. This can work only on WIN32 (combining - Proactor with Reactor's event loop). So I have put a #if defined - around the file. - -Fri Apr 16 18:09:44 1999 Irfan Pyarali <irfan@cs.wustl.edu> - - * tests/Conn_Test.cpp (server): If we timeout on accept(), it is - not an error, and should not be reported as one. Thanks to - Thomas Mehrkam <Thomas_Mehrkam@i-o.com> for pointing this out. - -Fri Apr 16 13:08:49 1999 David L. Levine <levine@cs.wustl.edu> - - * ace/OS.{h,cpp},ACE.{h,cpp}: (read_n,send_n,write_n): moved - from ACE to ACE_OS, so that the uses in OS.cpp could be localized. - The ACE versions are now just wrappers over the ACE_OS - versions, for backward compatibility. Thanks to - Erik Ivanenko <erik.ivanenko@utoronto.ca> for reporting - that OS.cpp used a couple of the ACE static member functions. - -Fri Apr 16 10:13:14 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/rules.nested.GNU,ACE-INSTALL.html: added - support for recursive make invocations with Windows NT's shell. - Thanks to Dave Steele <Dave_Steele@Mitel.COM> for diagnosing the - problem, and Tres Seaver <tseaver@palladion.com> for providing - the wild NT FOR command. Check it out, Darrell. - -Fri Apr 16 07:52:11 1999 Darrell Brunsch <brunsch@cs.wustl.edu> - - * ace/High_Res_Timer.cpp: Fixed my previous fix with - calibrate being called on Windows NT when a high-res - timer of 1 Mhz was being found. Calibrate should not - be called on NT now. Thanks to Zoran Ivanovic - <zorani@pathcom.com> for noticing this. - - * ace/Profile_Timer.cpp: Integrated the patch from Andy - Marchewka <AndyM@who.net> for returning rusage information - from elapsed_time. - -Thu Apr 15 16:50:21 1999 David L. Levine <levine@cs.wustl.edu> - - * ACE-INSTALL.html: updated discussion of "dir unexpected" - error when building for VxWorks on NT. Thanks to - Dave Steele <Dave_Steele@Mitel.COM> for providing the - insight. - -Thu Apr 15 14:04:36 1999 David L. Levine <levine@cs.wustl.edu> - - * include/makeinclude/rules.lib.GNU: don't call $(RANLIB) - if it is null. - - * include/makeinclude/platform_vxworks5.x_{g++,ghs}.GNU: removed - RANLIB definition, so that it will be null. This prevents - calling "true" on NT hosts. - - Thanks to Thomas Mehrkam <Thomas_Mehrkam@i-o.com> for reporting - that his NT host doesn't have "true". - -Thu Apr 15 14:00:00 1999 Chris Gill <cdgill@cs.wustl.edu> - - * ace/OS.{h, i}: simplified the conditional compilation logic - for a couple of the pSOS-specific ACE_OS methods, removed a - handful of marker comments, added a conditional compilation - branch for pid_t declaration. - Wed Apr 14 22:40:08 1999 David L. Levine <levine@cs.wustl.edu> * tests/run_tests.sh: call run_tests.check via sh, to support platforms that don't have /bin/sh. Thanks to Thomas Mehrkam - <Thomas_Mehrkam@i-o.com> for reporting that his NT host doesn't - have /bin/sh. + <Thomas_Mehrkam@i-o.com> for reporting that it's necessary on + his NT host. Wed Apr 14 18:56:00 1999 Chris Gill <cdgill@cs.wustl.edu> @@ -2140,8 +11,7 @@ Wed Apr 14 18:56:00 1999 Chris Gill <cdgill@cs.wustl.edu> ace/config-psos-diab-ppc.h include/makeinclude/platform_psos_diab_ppc.GNU: changes to adapt the pSOS Diab 68k/MIPS port to PPC. Many thanks to Sarmeesha - Reddy <sreddy@tr.comm.mot.com> and Steven Tine - <stevent@tr.comm.mot.com> for contributing these changes. + Reddy <sreddy@tr.comm.mot.com> for contributing these changes. Wed Apr 14 17:06:22 1999 David L. Levine <levine@cs.wustl.edu> @@ -2213,6 +83,7 @@ Wed Apr 14 02:34:13 1999 Alexander Babu Arulanthu <alex@cs.wustl.edu> defined to SIGRTMIN and SIGRTMAX respectively, for POSIX4 platforms. For others, they are 0. + Tue Apr 13 15:29:37 1999 David L. Levine <levine@cs.wustl.edu> * ace/Object_Manager.cpp: moved ACE_Service_Config signal |